Orchestra + Clover

Orchestra is an AI-powered project management system that turns a project idea into a dependency-locked engineering roadmap, auto-assigns tasks to team members based on skill profiles, and tracks progress across GitHub and Discord. Clover is the conversational assistant that answers natural-language questions about your project using RAG search and a Neo4j knowledge graph.

What it does

1. Blueprint generation — describe your project, provide your tech stack and team members, and Orchestra generates a full task roadmap with priorities, deadlines, dependencies, and assignments using Gemini
2. Smart assignment — tasks are assigned based on each team member's skill profile, scoped only to the project team
3. Knowledge graph — tasks, developers, and skills are stored in Neo4j with relationships, powering dependency tracking and skill gap detection
4. Clover — ask questions like "what is blocked?" or "who is working on the API?" and get answers grounded in live task data and recent GitHub/Discord activity
5. Activity tracking — GitHub commits and Discord messages are ingested, normalized, and linked to roadmap tasks in real time

Team

Name	Role	GitHub
Mitaali Singh	Lead · PM · AI Developer	@mitaalisingh
Naman Gupta	Knowledge Graph Engineer	@Naman-GG
Arnav Tripathi	Infrastructure Engineer	@ArnavXT
Sarvagya Prakash	Data Pipeline Engineer	@SarvagyaPrakash
Prince Negi	Interactive Canvas Specialist	@PrinceNegi
Isha Mahadev	Interface Developer	@IshaMahadev

Repos

Repo	Description
orchestra-ai	AI server — blueprint generation, task assignment, Clover assistant, knowledge graph, semantic search
orchestra-backend	Backend API — auth, event ingestion from GitHub and Discord, WebSocket feed, Postgres
orchestra-frontend	Web app — project creation, workflow canvas, kanban board, Clover chat, calendar

Architecture

User fills in project form (name, description, tech stack, members)
        |
        v
Frontend → Backend proxy → POST /blueprint (AI Server)
        |
        v
Gemini generates tasks with IDs, priorities, deadlines, dependencies, assignments, summary
        |
        ├── Tasks saved to Neo4j (knowledge graph)
        ├── Tasks saved to Postgres (via POST /tasks)
        └── Project saved to Postgres (via POST /projects)
        |
        v
Frontend displays workflow canvas (ReactFlow) + kanban board
        |
        v
GitHub/Discord events ingested by backend → linked to tasks by AI server
        |
        v
Clover answers questions using task data + graph + live events (RAG + Gemini)

Security: Frontend never calls the AI server directly. All AI server endpoints are protected by INTERNAL_API_KEY — the backend holds the key and proxies requests. This keeps the key out of the browser.

API endpoints (AI Server)

Method	Endpoint	Description
POST	/blueprint	Generate roadmap from project details
POST	/assign	Assign tasks to team based on skills
POST	/clover	Ask a project question (RAG + Gemini)
GET	/tasks	Get all tasks from Neo4j
GET	/graph	Get knowledge graph (ReactFlow format)
GET	/search	Semantic task search
GET	/standup	Generate per-person standup messages
GET	/replan	Suggest reassignment for blocked tasks
GET	/skill-gap	Detect tasks with no skilled assignee
POST	/onboarding	Infer skills from GitHub and add developer to graph
POST	/commit-intel	Link GitHub/Discord events to roadmap tasks
GET	/team	Get all team members and skills
POST	/team/manual	Manually set a developer's skills

Live URLs

Service	URL
AI Server	https://orchestra-ai-36zm.onrender.com
Backend	https://orchestra-backend-30fy.onrender.com
API Docs	https://orchestra-ai-36zm.onrender.com/docs

Stack

Python · FastAPI · Gemini API · Neo4j · ChromaDB · PostgreSQL · React · ReactFlow · Render